Local keyword research for location pages: choosing which towns and services deserve a page

Local keyword research for location pages is a decision about which town-and-service combinations have real demand AND enough distinct content behind them to deserve a page. The goal is restraint, not coverage. Most failed location-page programmes are not under-researched, they are over-built: a tool returned 300 town-and-service combinations and someone published all 300, including the 250 that had no demand and nothing unique to say. This guide covers how to map services against towns, how to tell real demand from a tool artefact, and how to spot the combinations that would only ever produce a thin near-duplicate before you waste a page on them.

This page owns the upstream decision: does this town-by-service combination have demand at all, and could the page behind it ever be distinct? Once you have a shortlist of combinations worth building, the companion guide on how many location pages to build and in what order owns sequencing and volume. The two do not overlap: research decides what is eligible, the build plan decides what ships first.

What is the difference between explicit and implicit local queries?

An explicit local query names the place. “Boiler repair Uckfield” and “emergency electrician Lewes” are explicit: the searcher has typed the town, so the intent to find a local provider is on the page in plain words. An implicit local query carries local intent without naming a place. “Boiler repair near me”, “emergency electrician open now”, or a bare “boiler repair” typed on a phone in Uckfield are all implicit: Google infers the location from the device, not the words.

This distinction decides what a location page is for. A location page earns its keep against explicit queries, where the searcher has told Google the town and Google can match a page that is genuinely about that town. Implicit “near me” demand is mostly satisfied by your Google Business Profile and the map pack, not by a dedicated page, which is why chasing a separate page for every “near me” variant is usually wasted effort. If you want the longer version of why proximity-driven searches behave differently, the guide on near-me searches for local businesses covers it.

• Explicit query: names the town. A dedicated, distinct location page can rank for it. This is what you are researching for.
• Implicit query: “near me”, “open now”, or place inferred from device. Handled mainly by your Business Profile and proximity, not a separate page.
• Practical rule: research explicit town-plus-service phrases. Do not build a page hoping to capture an implicit query a profile already serves.

How do you map services against towns to find real combinations?

Seed-by-location mapping is a grid, not a keyword dump. You list your genuine services down one axis and the towns you actually serve across the other, then you test each cell for two things in order: is there explicit demand, and could the page be distinct. A cell only survives if both answers are yes. The method matters more than any tool, because the tool will happily multiply 12 services by 40 towns into 480 cells and none of that arithmetic tells you which cells a real customer ever searches.

Work the grid in this order so you spend research effort where it pays:

1. Seed the services honestly. List only services you actually deliver and would quote on. A plumber who does not fit bathrooms should not have a “bathroom installation” row, no matter how the keyword tool tempts.
2. List the towns you serve, not the towns you can reach. A van that could drive to a town is not the same as a business with anything specific to say about working there. Reach is not demand and it is not distinctness.
3. Test each cell for explicit demand. Does anyone search “[service] [town]” as an explicit phrase. A few real searches a month is enough for a trade. Zero, across every variant, removes the cell.
4. Test each surviving cell for distinct content. Could this page carry something true and specific to that town that no other page on your site could carry. If the honest answer is “only the town name changes”, the cell fails here.
5. Keep only cells that pass both tests. The survivors are your candidate pages. Everything else is a row in a spreadsheet, not a page on your site.

A grid of 480 cells routinely collapses to 30 or 40 real candidates once you apply both tests honestly. That collapse is the research working, not failing. For what actually fills a surviving page so it is not just a swapped town name, see location page content ideas, and when you are taking on a town for the first time, how to build a service-area page for a new town walks through the page itself.

How do you judge demand against saturation?

Demand for a trade is rarely about large numbers. “Boiler repair Uckfield” might show a tiny monthly volume in a keyword tool and still be worth a page, because one boiler repair is a real job and the searcher has high intent. The trap is not low volume, it is zero distinct demand dressed up as opportunity. A tool that reports volume for “boiler repair [village of 400 people]” is often showing you a regional rounding figure, not real local searches.

Read the live results, not just the tool. Search the explicit phrase and look at who already ranks. If the first page is established local firms with detailed, genuinely local pages, that is a saturated cut you may not win cheaply. If the first page is thin aggregator listings and directories with no real local provider, that is an opening worth a distinct page. Saturation is about the quality of what ranks, not the count of competitors.

• Low volume, high intent: usually worth a page for a trade. One job covers it.
• Volume on a tiny place: treat as suspect. Check whether it is a regional figure attributed to the nearest named place.
• Saturated by strong local pages: a thin page will not break in. Either go deeper than the incumbents or skip the cut.
• Saturated by directories only: a genuinely local, specific page can outrank them. Worth building.

Keyword-tool mechanics are deliberately a small part of this. The tools are good at returning phrases and rough volumes and bad at judging whether a page can be distinct, which is the decision that actually protects you. Spend your time on the live SERP and the distinctness test, not on tool walkthroughs.

When does a neighbourhood page stop being worth it?

A neighbourhood page stops being worth it at the point where it can no longer say anything true that the parent town page does not already say. Town-level pages usually clear the distinctness bar because towns differ in real ways: different jobs, different building stock, different access, different demand. Neighbourhood and suburb pages start failing it fast, because two suburbs four streets apart almost never differ in anything a customer cares about.

Apply this test to any neighbourhood-level candidate before you build it:

• The substitution test: draft the page in your head, then swap the neighbourhood name for the next one over. If every other sentence still reads as true, the page has no distinct content and should not exist as a separate page.
• The explicit-demand test: does anyone actually search “[service] [neighbourhood]” as an explicit phrase. Most neighbourhoods generate implicit “near me” intent, not explicit named searches, and that intent belongs to your Business Profile.
• The genuine-difference test: can you name one concrete thing about working in this neighbourhood that is not true of the wider town. A specific access constraint, a common local job, a building type. If you cannot, fold it into the town page.

For most trades the answer is one strong town page, not five thin neighbourhood pages under it. When a town legitimately splits, a city the size of Brighton with distinct named areas that people search by name, you build those few that pass all three tests and fold the rest. The decision on overall volume and order, once eligibility is settled, sits with the how-many-location-pages guide.

How do you spot a combination that would only ever be a thin near-duplicate?

A thin near-duplicate is a page whose only variable is the place name. You can spot one before you build it. The signs are consistent across trades, and every one of them means the same thing: there is nothing town-specific to write, so the page would be filler with a town pasted in.

• You can only describe the page as “the [town] version of the [service] page”. That is a duplicate with a find-and-replace, not a new page.
• You have no real proof tied to the town: no job you have done there, no photo, no review, no specific local detail. Generic claims about “the [town] area” are not proof.
• The only difference between this candidate and the last one is the noun. If your drafts differ by a single word, they are one page.
• You would have to invent the local content to fill the page. The moment you are inventing, stop. Invented locality is worse than no page.
• The town appears nowhere in your actual work history and you have no concrete plan to serve it. Reach without evidence is not a page.

Publishing combinations with no distinct content behind them is exactly what Google’s spam policies describe as scaled content abuse, which Google defines as “many pages are generated for the primary purpose of manipulating search rankings and not helping users”, a practice it acts on “no matter how it’s created” (Google Search Central, Spam policies for Google web search, accessed 11 August 2026). Pages built only to rank for “specific, similar search queries” that then route people to a more useful page elsewhere are the doorway pattern Google describes under doorway abuse in the same policy. Neither problem is about volume. A site with twelve genuine, distinct town pages is fine. A site with twelve near-identical ones is at risk. The guides on doorway pages and the scaled-content policy go deeper, and duplicate content across location pages covers the near-duplicate failure in detail.

How does Townsmith enforce the demand-versus-distinctness decision?

The judgement this whole guide describes is the exact decision Townsmith’s quality score and hold-back behaviour exist to enforce on the page itself. Research decides which combinations are eligible. The quality score then checks, page by page, whether the content you actually wrote is distinct enough to publish. A page that is a near-duplicate, that is mostly a swapped town name, or that has no real local substance scores low and is held back rather than published. That is the substitution test and the thin-near-duplicate signs from this guide, applied automatically at build time.

The free plugin does this entirely on your own server with no AI and no external calls. It builds real, editable location pages, scores each one from 0 to 100, holds back the thin ones, and outputs the schema. It is the enforcement layer for the restraint this guide argues for, not a way to mass-produce the pages it warns against. You can see what the score measures in the quality-score documentation, and what the engine builds on the features page. It coexists with Rank Math, Yoast, AIOSEO and SEOPress rather than replacing your SEO plugin.

None of this guarantees rankings or compliance, and nothing can. What it does is make the demand-versus-distinctness decision unavoidable: the 250 cells that should have failed the grid will not quietly slip through, because a page that has nothing town-specific to say scores low and is held back rather than published. The decision in this guide stops being advice you have to remember and becomes a gate the build has to pass.

Common questions about local keyword research for location pages

Do I need a separate keyword tool to do this well?

No. A tool helps you gather candidate phrases and rough volumes, but the decisions that matter, explicit versus implicit intent, real demand versus a regional artefact, and whether a page can be distinct, come from reading the live results and applying the substitution test. The tool cannot judge distinctness, and distinctness is what protects you. Keep tool use light and spend the time on the SERP and the grid.

Is low search volume a reason not to build a town page?

Not on its own. For a trade, one boiler repair or one emergency call-out is a real job, so a page with genuine local intent and low reported volume can still be worth building. The reasons to skip a page are no explicit demand at all and no distinct content to put on it, not a small number in a volume column.

Should I build a page for every town my van can reach?

No. Reach is not demand and it is not distinctness. Build pages for towns where people explicitly search for your service and where you have something true and specific to say about working there. For the rest, a single well-built service-area page that names the towns you cover, set out in the local landing page checklist, serves them without spawning thin near-duplicates.

How is this different from the how-many-location-pages guide?

This guide decides which town-and-service combinations are eligible at all: whether there is explicit demand and whether the page could ever be distinct. The how-many-location-pages guide takes the eligible shortlist and decides how many to build and in what order. Use this page to build the shortlist, then that one to sequence it.

Is there a special AI or GEO step for getting these pages cited?

No. There is no AI-only schema, and no llms.txt or chunking trick that gets a location page into AI Overviews or AI Mode. A page that is genuinely about a town, indexed and snippet-eligible, is what gets cited. The same demand-and-distinctness discipline in this guide is the optimisation. Structured data helps machines read a page that already deserves to rank, it does not rescue one that does not.

Figures and policy references current as of 11 August 2026. Written by Stephen Evans. Primary source: Spam policies for Google web search, Google Search Central, accessed 11 August 2026.